Diode reverse voltage w/center tapped transformer

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have a transformer with a 660v center tapped secondary, so 330v-0-330v across the winding. With a full wave rectifier layout as per the attachment below, is the reverse voltage seen at the diode 330v, or 660v?
 

Attachments

  • Capture.JPG
    Capture.JPG
    19 KB · Views: 173
This is a full wave rectifier, not a bridge.

Diode will see twice the peak voltage, which in this configuration is 1.4142*330VAC , so twice 467VDC = 934V

A nominally 1000V diode, such as popular 1N4007 customarily stands this, because of course it's made with some extra safety margin.
 
Member
Joined 2011
Paid Member
I'd suggest a robust surge suppressor after the mains fuse and before the transformer primary. For example this TransZorb is rated 1500 watts, costs less than a dollar, and protects your 1800 volt diodes in the secondary circuit by popping the mains fuse. Connect a 22 kiloampere MOV such as this one in parallel, for extra piece of mind. A voltage surge is clamped, the (enormous) clamping current blows the fuse, which protects the downstream electronics like your fragile 1.8 kilovolt diodes.

_
 
Last edited:
I'd suggest using 2x 1N4007 or UF4007 in series for each diode (assuming current rating is fine), without any balancing caps or resistors. Although 1N4007 is 1kV PIV rated, and may have some margin, the AC mains could be 10-15% off the rated primary voltage of your transformer, as you haven't really identified that side of the equation.

Using 2 series diodes of the same type/batch should give you sufficient extra PIV (not 2x PIV, but certainly a good percentage of that). And it also reduces the off-state diode capacitance, and so alleviates noise egress from the PT side of the diodes.

A compromise on the brickwall primary side transient protection suggested by Mark would be a smaller MOV alone placed across the primary winding. Such a MOV would clip any primary winding induced voltage transient induced by a power switch turning off. If you feel as if your equipment needs protection from high energy transients/surges coming in from external mains AC (eg. lightning strike, or a HV line falling on an LV line), then I suggest that protection is best located at your switchboard, or in a local power board that also protects other audio gear as well.
 
All my equipment is behind a mains filtering unit so i should be fairly well protected against spikes.

I am very tempted to use a rectifier tube instead, i tried an 83 mercury rectifier and it seems to work just fine, datasheet specs 1550v inverse voltage, 550v rms peak per plate limit, and 100ma is less than half its rated current. Unless i am missing something it seems ideal for the job, and looks pretty too :)
 
Yes i did, was not a trivial task either as i ended up having to use a 2uf filter cap and a large resistor to keep things in check. Gone off the idea of rectifier tubes now, too many limitations and downsides without any positives, other than looking pretty. Fair to say the novelty has worn off.

Back to looking as solid state options, can anybody explain the pros and cons of capacitance and reverse current when looking at diodes and rectifiers? I am assuming the lower the two, the better.

Cheers,

Mark
 
Member
Joined 2011
Paid Member
Some people solder a capacitor directly across the terminals of each diode in a bridge rectifier. Four diodes, four capacitors. I would imagine those folks don't care about diode capacitance at all, since Cacross >> Cdiode.

For those who use a CRC snubber between the transformer and the rectifier diodes, it's important to know the value of Cdiode, in order to set the snubber Cx >> Cdiode. Fortunately the vast majority of diodes are less than 1.5nF and so the "recommend starting point" Cx=10nF, is indeed much greater than Cdiode. Whew, a design decision avoided! In the rare cases when Cdiode is more than 1.5nF that's not the end of the world, it just means you need to select a larger Cx and a larger Cs. Google for Quasimodo test jig to learn more.

Reverse current would not be an issue unless it is a non-negligible fraction of the power supply output load current. I wouldn't pay attention to the reverse current rating unless Ireverse > (Ioutput / 50). But this has never happened, at least, not to me.
 
For the diodes i am looking at we are talking 8-20pf so i assume its not worth worrying about then.

I usually do fit snubbers across the secondary as i built my own cheapomodo for testing. In this case i decided against it due to the voltages involved, having enough trouble getting diodes to work never mind snubbers :)
 
For simple power supplies, ss diodes are equally simple to use.

For audio applications, where people spend extra effort to alleviate 'noise and spikes' that get past the power supply and can be 'seen' on the output signal then that requires a little more attention to detail.

Diode current is meant to stop when voltage reverses, but diode capacitance and reverse current let through 'noise' currents that then has a chance to get past the power supply filtering and out to B+ (and then eg. via output transformer capacitance to speaker leads).

Reverse current can happen at the moment of reverse biasing, in which cast the faster diodes (such as UF4007) are better. Reverse current that is forced by reverse bias voltage is very temperature and voltage dependant.

Diode capacitance lets through the dV/dt character of the transformer winding voltage. Unfortunately, winding voltage is not always nicely sinusoidal, but includes many harmonics from the mains AC, as well as a substantial step in voltage due to current stopping at diode turn off. Diode capacitance increases with diode current rating, so using the lowest current rated diode helps. Using two diodes in series, halves the total diode capacitance, so that helps. Using a secondary winding snubber helps by trying to smooth out the voltage at the time of diode turn off.

SS diodes induce a 'harder' turn off than valve diodes, and so often don't exhibit the noise/spikes that are often observed at the time of diode turn-off when looking at a signal waveform.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.